Research On Video Transmission Quality Evaluation And Encoder Adaptation Scheme

Along with the rapid development in video compression standards and network transmission technologies,video streaming application has faced to a situation where the end-user expects high quality of experience(QoE)available.QoE,defined by ITU-T,is the ultimate measure of user satisfaction to be maximized in multimedia application.Increasing focus has gradually turned from Quality of Service(QoS)awareness to end viewer's Quality of Experience(QoE)awareness for resource allocation and network optimization.In general,quality of online video is affected not only by application-layer QoS parameter caused by encoder compression,but also by network-layer QoS parameter.Video Streaming has become one of the most popular applications over wireless networks.The emerging High Efficiency Video Coding(HEVC/x265)brings significant higher compression efficiency compared with H.264/AVC.It can bring up 50% bandwidth saving with negligible loss in video quality.This bandwidth saving makes HEVC/H.265 an attractive compression method for the delivery of video content over the resource-constrained network.However,it potentially leads to higher sensitivity to the network impairment,and thereby affects the users experience awareness.Thus,a QoE derivation approach focused on the joint consideration of information collected along different layers is desired.At the same time,for multi-user video streaming services,videos are encoded and transmitted through a bandwidth-limited wireless network.Therefore,proper encoder settings for videos are required to satisfy the total channel rate constraint and improve the user satisfaction.This thesis presents a systematic research on the QoE evaluation methods.We classify them into three categories and summarize the most representative QoE evaluation techniques.We also simulate various real-time and non real-time applications using the network simulation software Opnet.By experimental results,we obtain their performance curves and analyze the QoS parameters.Besides,this thesis mainly analyzes the QoE performance of HEVC-based video transmission to develop a unified QoE prediction framework for HEVC encoded video streaming.The impact of network impairment is investigated by defining different packet loss rates and the impact of HEVC encoder is fully studied by selecting the most relevant feature from encoding process.Moreover,the impact of video content is formulated using the average bits per pixel and depth information.Finally the impacts of network impairment,HEVC encoder and video content are incorporated into the QoE metric as multiplicative factors.Results show that the correlation accuracy between actual QoE values and predicted QoE values reaches 93.4% and outperforms other existing QoE metrics.This thesis also proposes a QoE-driven HEVC encoder adaptation scheme,aiming to maximize QoE of all the users for wireless networks.Firstly,the influence of HEVC encoder on video streaming is investigated to generate an encoder parameter model.Then,the effect of network impairment together with HEVC encoder is taken into account to derive a subjective QoE prediction model.Afterwards,we formulate a QoE-maximized encoder adaptation with total channel rate constraint into an optimization problem based on the obtained encoder parameter model and the subjective QoE model.The problem is solved and incorporated into an online encoder adaptation scheme.Compared with conventional rate allocation methods,the proposed QoE-driven encoder adaptation scheme achieves significant QoE gains independent of wireless network channel rate and video content.Simulation results demonstrate that the proposed encoder adaptation scheme can improve the QoE by 31.4% with fixed number of users or increase the number of satisfied users by 35.5% with variable number of users.